Predictor relevance and extramarital affairs

We revisit Fair's (1978) 'theory of extramarital affairs' using robust nonparametric methods developed for the analysis of categorical data. We find evidence suggesting that the number of years married is not a relevant predictor of the propensity to engage in extramarital affairs having controlled for other factors. This finding runs counter to the prevailing wisdom gleaned from misspecified parametric models. Copyright © 2004 John Wiley & Sons, Ltd.